Determination of the optimal model complexity for process optimisation of biofilm reactors

Biofilm systems, e.g. trickling filters, rotating biological contactors and submerged filters, have - second to activated sludge systems - always been very popular in wastewater treatment processes. The growth of a biofilm on a carrier material has the advantage of allowing a high biomass density. These systems therefore only need a small reactor volume. In addition a lot of microniches exist in a biofilm, encouraging a high species diversity. Biofilm configurations are also more resistant to toxic pulses and fluctuating influent characteristics due to the short hydraulic retention time in the system. It is also true, though, that the growth of biomass on a carrier material is difficult to control. For example the oxygen transfer to the biofilm can hardly be manipulated and it is almost impossible to control the biofilm thickness. This can lead to blocking and to the loss of biofilm by sloughing.

More modern process configurations, like fluidized bed and granular activated carbon filters, have a better performance in this respect. These configurations use particles as carrier material. The biofilm is usually thin, because the thickness is limited by important friction forces. However, these reactors have a high specific surface area and therefore guarantee a high biological activity. In the case of the activated carbon filter the adsorption of substrate molecules to the activated carbon also plays an important role.

For all these configurations, the use of modelling and simulation can be a big help for the further understanding of the processes taking place in the reactor, certainly for the kinetics and the interaction between several processes. Reliable models can - via optimisation studies - also lead to a better design, a better use and a better control of the biofilm reactor.

Up to now, two types of biofilm models have been developed and used. The first type, developed in the seventies and eighties, uses a steady-state or a pseudo steady-state approach. The biofilm is represented by a flat and homogeneous structure of constant depht. Growth and decay of biomass are not considered explicitly and the substrate removal is calculated from a steady-state concentration profile through the biofilm. Later, several extensions have been added to cope with the presence of different species and substrates in the system. The dynamics of these extensions are limited to the calculation of the biofilm thickness by an algebraic equation, while the fast dynamics are still neglected.

In the second half of the eighties, full dynamic models of the biofilm were developed. These models inevitably lead to the use of several empirical relations to describe the process dynamics that are not fully understood. Moreover, they require an extensive calibration because they have a considerable number of parameters.

We therefore can ask the question which model complexity is optimal for the description of biofilm systems. In other words, how far can we simplify a model without damaging its capacity to describe the process accurately. The fields where biofilm models are used impose a limit to their complexity because the calibration and the calculation intensity have to be feasible. The calibration has to be possible using input and output data of full scale plants.

Project duration: 1997-10-01 - 2001-09-30

Project home page


Henk Vanhooren
Department of Applied Mathematics, Biometrics and Process Control
Coupure Links 653
9000 Gent
Tel: +32(0)9 264.59.35
Fax: +32(0)9 223.49.41

Last update: 01 december 2008,

  zoek print log in nederlands